DeepSeek V4.1 Flash vs GPT-5.6 Luna vs LongCat-2.5-Preview
No independent capability benchmark covers every model here yet. The score uses inputs & features and context window.
Price in rating: Excluded · Include price
DeepSeek
DeepSeek V4.1 Flash
62/100- ECI157.0
- Price$0.232 / $1.08
- Context1M
OpenAI
GPT-5.6 Luna
74/100- ECI156.7
- Price$0.20 / $1.20
- Context1.05M
Meituan (LongCat)
LongCat-2.5-Preview
62/100- ECI—
- Price$0.30 / $1.20
- Context1M
Capability evidence is incomplete
There is not enough shared capability evidence to name an overall winner. The scores (GPT-5.6 Luna 74/100, DeepSeek V4.1 Flash 62/100, LongCat-2.5-Preview 62/100) compare the available specifications only. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceDeepSeek V4.1 FlashDeepSeek V4.1 Flash $0.444 · GPT-5.6 Luna $0.45 · LongCat-2.5-Preview $0.525 per 1M tokens (3:1 blend)
- Longest contextGPT-5.6 LunaGPT-5.6 Luna 1,050,000 · DeepSeek V4.1 Flash 1,000,000 · LongCat-2.5-Preview 1,000,000 tokens
- Widest inputsGPT-5.6 LunaDeepSeek V4.1 Flash: Text, Images · GPT-5.6 Luna: Text, Images, PDFs · LongCat-2.5-Preview: Text, Images
- Self-hostingDeepSeek V4.1 FlashPublishes downloadable weights (MIT)
| Measure | Weight | DeepSeek V4.1 Flash | GPT-5.6 Luna | LongCat-2.5-Preview |
|---|---|---|---|---|
| Inputs & features | 60% | 60 | 80 | 60 |
| Context window | 40% | 64 | 65 | 64 |
| Specs only | 100% | 62/100 | 74/100 | 62/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 157.0 (best) | 156.7 | — |
| ECI rank | #34 of 158 (best) | #36 of 158 | — |
| GPQA DiamondGraduate-level science questions | 89.8% | 91.6% (best) | — |
| FrontierMath Tiers 1–3Research-level mathematics | 67.4% | 82.1% (best) | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 98.3% | 98.3% | — |
| SimpleQA VerifiedShort factual questions | 42.9% (best) | 41.0% | — |
| Price per million tokens | |||
| Input | $0.232 | $0.20 (best) | $0.30 |
| Output | $1.08 (best) | $1.20 | $1.20 |
| Cached input | — | $0.02 | $0.006 (best) |
| Blended (3:1) | $0.444 (best) | $0.45 | $0.525 |
| Long-context rate | Same rate | Over 272K: $0.40 / $1.80 | Same rate |
| Price source | Median of 82 providers | Official OpenAI API | Official LongCat API |
| Limits | |||
| Context window | 1,000,000 tokens | 1,050,000 tokens (best) | 1,000,000 tokens |
| Max output | 384,000 tokens (best) | 128,000 tokens | 131,072 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | No | Yes | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | Yeslow · medium · high · xhigh · max | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | OpenMIT | Proprietary | Proprietary |
| API model ID | — | gpt-5.6-luna | LongCat-2.5-Preview |
| API providers | 86 (best) | 57 | 3 |
| Released | Sep 10, 2026 | Jul 9, 2026 | Sep 25, 2026 |
| Knowledge cutoff | May 2025 | Feb 16, 2026 | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
DeepSeek V4.1 Flash$4.47
GPT-5.6 Luna$4.40
LongCat-2.5-Preview$5.40
Which should you choose?
Which is better: DeepSeek V4.1 Flash, GPT-5.6 Luna or LongCat-2.5-Preview?
There is not enough shared capability evidence to name an overall winner. The scores (GPT-5.6 Luna 74/100, DeepSeek V4.1 Flash 62/100, LongCat-2.5-Preview 62/100) compare the available specifications only. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, DeepSeek V4.1 Flash, GPT-5.6 Luna or LongCat-2.5-Preview?
DeepSeek V4.1 Flash is cheaper at $0.232 input / $1.08 output per million tokens (median across 82 API providers). GPT-5.6 Luna costs $0.20 input / $1.20 output per million tokens (official OpenAI API price); LongCat-2.5-Preview costs $0.30 input / $1.20 output per million tokens (official LongCat API price). At a typical mix of three input tokens to one output token, that is $0.444 per million tokens for DeepSeek V4.1 Flash versus $0.45 for GPT-5.6 Luna (1× as much) and $0.525 for LongCat-2.5-Preview (1.2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. DeepSeek V4.1 Flash has an ECI of 157.0, GPT-5.6 Luna has an ECI of 156.7 and LongCat-2.5-Preview has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for DeepSeek V4.1 Flash, GPT-5.6 Luna and LongCat-2.5-Preview yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
GPT-5.6 Luna has the largest context window at 1,050,000 tokens, against 1,000,000 for DeepSeek V4.1 Flash and 1,000,000 for LongCat-2.5-Preview. Maximum output per response: DeepSeek V4.1 Flash up to 384,000, GPT-5.6 Luna up to 128,000, LongCat-2.5-Preview up to 131,072 tokens.
Which can read images, PDFs, audio or video?
DeepSeek V4.1 Flash accepts text and images; GPT-5.6 Luna accepts text, images and PDFs; LongCat-2.5-Preview accepts text and images. GPT-5.6 Luna handles the widest range of inputs.
Are any of these open source?
DeepSeek V4.1 Flash publishes its weights (MIT) and can be self-hosted; GPT-5.6 Luna and LongCat-2.5-Preview is proprietary.
Which is newer?
LongCat-2.5-Preview is the newest, released Sep 25, 2026. DeepSeek V4.1 Flash came out Sep 10, 2026; GPT-5.6 Luna came out Jul 9, 2026. Knowledge cutoff: DeepSeek V4.1 Flash May 2025, GPT-5.6 Luna Feb 16, 2026.
How do you decide the winner?
This rating uses inputs & features (60%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (40%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Nov 18, 2026.